Vulnerable to landslides Artificial intelligence (AI) techniques can be useful in identifying areas that are more prone to landslides. A study published in October 2024 in the scientific journal Natural Hazards Research compared the effectiveness of five models based on machine learning algorithms to identify and predict the points in the municipality of São Sebastião that are most susceptible to this type of event. According to the article. By researchers from São Paulo State University (Unesp) and the National Center for Monitoring and Alerts for Natural Disasters (Cemaden), one of the models, Gradient Boosting, achieved 99.6% accuracy in mapping the areas most vulnerable to landslides. With almost identical performance, the Random Forest algorithm came in second place in the ranking prepared by the authors of the study.
The study covered a territory of just over
square vulnerable to landslides rcs data kilometers (km2) of the resort located on the northern coast of São Paulo, an area subject to heavy rainfall and soil movement coming from the slopes of the Serra do Mar mountain range. Between February 18 and 19, 2023, during Carnival, more than 600 millimeters (mm) of rain fell in São Sebastião, the equivalent of two months. There were landslides, house collapses, 2,400 people were left homeless and 64 lost their lives. To classify the performance of the algorithms, the results of the models were compared with maps of the region showing the areas most subject to this type of occurrence.
The algorithms calculate the risk of landslides
occurring in a given location by analyzing data on environmental factors associated with processes that influence soil stability. The main elements taken into account how to appear on google: 5 steps to promote your business are the degree of terrain inclination. Soil moisture, dissection (fragmentation) of the relief. And geomorphological parameters of the region. “Machine learning models allow the integration of several conditioning. Variables and provide a robust basis for creating susceptibility maps.” Says remote sensing specialist Enner Alcântara. From the Institute of Science and Technology (ICT) at Unesp, São José dos Campos campus, lead author of the study. “They make it possible to identify complex patterns that may not be evident in more traditional approaches.”
The Gradient Boosting algorithm has
a unique feature: it combines approaches from lack data several simpler models, specialized in each of the variables analyzed. This more integrated view indicates that slope. Land fragmentation and soil moisture index are the factors that most impact slope stability. “vulnerable to landslides Greater forest cover.
Based on the Gradient Boosting model
a landslide susceptibility map was generated in which points in São Sebastião were classified into four risk categories: low (74.6% of the municipality’s territory), moderate (15.8%), high (7.9%) and very high (1.7%).
Despite the predominance of areas of low susceptibility
Occupied by scars opened in the ground by past landslides. Ssuch as in the Serra do Mar State Park. Near Juqueí beach and in Vila Sahy. In the February 2023 tragedy, most of the deaths occurred in the latter location.